Effect of a rs485 damaged driver

[Reginald Jean Louis]

Hi everyone.

I have a bunch of rs485 transceivers arrange in a 4-wires bus topology (my
question is good for 2-wire/half duplex too). I want to know if a blown up
driver can affect the line by putting permanently the line in a high or low
state? If so, there is a way to prevent that?

Thanks you!
Reginald Louis

 

[Tim Shoppa]

I have a bunch of rs485 transceivers arrange in a 4-wires bus topology (my
question is good for 2-wire/half duplex too). I want to know if a blown up
driver can affect the line by putting permanently the line in a high or low
state? If so, there is a way to prevent that?

Receivers can latch up/blow up too, for example by shorting one or both
of their inputs to Vcc or ground.

The most common failure I've seen is shorted output drivers holding one
line high or low. Sometimes the system still continues to sort-of work
but not reliably at all.

I've seen some applications guard against this by putting 50-ohm-or-greater
resistors in series between each node and the bus wires. Noise immunity
is decreased but the idea is that a "good" driver will outvote a "bad" driver
that has only one of its outputs shorted to ground/Vcc. This doesn't really
help if the "bad" driver is just ignoring its tristate input and jabbering
all the time as it then (electrically) has just as much vote as a good
driver.

Some bus networks have receivers that can issue an alarm if one of the
bus lines is stuck high/low or they see meaningless jabber. Haven't
seen this for RS-485 although some software will issue somewhat meaningful
alarms when they cannot see their own transmissions .

Tim.

 

[Tim Wescott]

Receivers can latch up/blow up too, for example by shorting one or both
of their inputs to Vcc or ground.

The most common failure I've seen is shorted output drivers holding one
line high or low. Sometimes the system still continues to sort-of work
but not reliably at all.

I've seen some applications guard against this by putting 50-ohm-or-greater
resistors in series between each node and the bus wires. Noise immunity
is decreased but the idea is that a "good" driver will outvote a "bad" driver
that has only one of its outputs shorted to ground/Vcc. This doesn't really
help if the "bad" driver is just ignoring its tristate input and jabbering
all the time as it then (electrically) has just as much vote as a good
driver.

Some bus networks have receivers that can issue an alarm if one of the
bus lines is stuck high/low or they see meaningless jabber. Haven't
seen this for RS-485 although some software will issue somewhat meaningful
alarms when they cannot see their own transmissions .

Tim.

Asynchronous serial receivers can detect framing errors, and any
protocol should include at least a checksum on messages -- you can alarm
on either bad checksums, framing errors, or the line going silent.

 

[Reginald Jean Louis]

I know I can detect bad frame, I just wanted to know if I can prevent a
damaged driver/receiver to corrupt the bus line.

 

[Tim Shoppa]

Asynchronous serial receivers can detect framing errors, and any
protocol should include at least a checksum on messages -- you can alarm
on either bad checksums, framing errors, or the line going silent.

Yeah, but still RS-485 doesn't have all the features that Ethernet
over say 10Base2 has (In thinking of the OP's request, desirable
features would include collision detection and hubs that can autopartition
to isolate a jabbering leg).

I know, some of that can be layered on top of RS-485, but the
protocol itself is just electrical and doesn't talk about how to
layer that all on. (Besides the electrical part doesn't specify
collision detect even though a couple of manufacturers make parts
that support detection of fighting drivers.) Not to bad-mouth RS-485,
there's far worse electrical interfaces in everyday use!

Tim.